Abstract
For the purpose of understanding second-order scalar PDEs and their hydrodynamic integrability, we introduce G-structures that are induced on hypersurfaces of the space of symmetric matrices (interpreted as the fiber of second-order jet space) and are defined by non-degenerate scalar second-order-only (Hessian) PDEs in any number of variables. The fiber group is a conformal orthogonal group that acts on the space of independent variables, and it is a subgroup of the conformal orthogonal group for a semi-Riemannian metric that exists on the PDE. These G-structures are automatically compatible with the definition of hydrodynamic integrability, so they allow contact-invariant analysis of integrability via moving frames and the Cartan--Kaehler theorem. They directly generalize the GL(2)-structures that arise in the case of Hessian hyperbolic equations in three variables as well as several related geometries that appear in the literature on hydrodynamic integrability. Part I primarily discusses the motivation, the definition, and the solution to the equivalence problem, and Part II will discuss integrability in detail.
The Classification Problem for Hessian PDEs
In the study of partial differential equations (PDEs), a fundamental challenge is classification: determining when two seemingly different equations are equivalent under a change of variables, specifically contact transformations. While a complete classification of all second-order PDEs is an immense task, this work focuses on Hessian PDEs—equations where the function depends exclusively on the second-order derivatives (the Hessian matrix).
Historically, geometric classifications for these equations were often "extrinsic," viewing the PDE as a surface embedded within a larger jet space, or were restricted to specific dimensions (e.g., $n=2, 3, 4$). This paper addresses the lack of a general, "intrinsic" geometric language capable of handling Hessian PDEs in any dimension $n$ while naturally identifying "integrable" equations—those admitting exact solutions via hydrodynamic reductions.
Geometric Framework: $Od(G)$-Structures and Cartan’s Method
The author employs Cartan’s Method of Equivalence and the theory of $G$-structures to build this framework. The approach treats a Hessian PDE as a hypersurface within the space of symmetric matrices, which represents the fiber of the second-order jet space.
Central to this framework is the introduction of $Od(G)$-structures. Here, the "fiber group" is a conformal orthogonal group acting on the independent variables of the PDE. By utilizing the method of moving frames, the author "prolongs" the structure to find a canonical coframe. This process yields structure equations—differential identities that encapsulate the geometry of the PDE.
To manage the complexity for computational applications, elementary representation theory is used to decompose the "torsion" and "curvature" of these structures into irreducible submodules. This identifies the core local invariants—the "DNA"—of the PDE.
Key Insights: The Veronese Connection and Intrinsic Geometry
A significant insight is the deep link between the geometry of these PDEs and the Veronese variety (the set of rank-one symmetric matrices). The interaction between the PDE's "symbol" and this variety is a primary determinant of its integrability.
By shifting to an intrinsic $G$-structure, the classification becomes contact-invariant. This ensures that if a PDE is transformed into a new coordinate system, its geometric invariants remain unchanged. The paper culminates in a Structure Theorem (Theorem 2.11), providing explicit differential equations for the invariants of any "embeddable" Hessian PDE. This serves as a roadmap for classification across any dimension. Furthermore, a "classifying space" $\mathbf{K}_\Phi$ is defined, where the leaves of a singular foliation represent equivalence classes of PDEs.
Results and Dimensional Consistency
The author successfully solves the "equivalence problem" for these structures through the First and Second Fundamental Lemmas, identifying which parts of the torsion are essential invariants and which can be absorbed.
The framework demonstrates robust consistency with established results in lower dimensions:
- For $n=3$, it recovers the $GL(2)$ geometry used in previous studies of integrable systems.
- For $n=4$, it aligns with the geometry of symplectic Monge–Ampère equations.
Additionally, the paper establishes existence and uniqueness results: for any valid point in the classifying space, there exists a corresponding analytic Hessian PDE that is locally unique up to contact transformation.
Limitations and Future Directions
While powerful, the current framework is strictly limited to Hessian PDEs. Although some evidence suggests this may provide a "coarse" classification for all second-order PDEs, a full general classification remains a future goal. The results also assume a "non-degenerate symbol"; singular or highly degenerate PDEs would require distinct geometric treatments.
Finally, this first part focuses on the foundational setup and the equivalence problem. The detailed exploration of hydrodynamic integrability—the property that makes these equations mathematically and physically unique—is reserved for Part II of this series. As the dimension $n$ increases, the number of invariants grows rapidly, making the explicit identification of the "moduli space" for $n \ge 3$ a significant ongoing computational challenge.
Summary updated